2025-05-16

Tech Leaders on Bitcoin, AI, and the New Global Power Shift w/ Salim Ismail & Dave Blundin | EP #172

Peter Diamandis, Salim Ismail, and Dave Blundin cover a week dominated by Bitcoin's return above $100K, Michael Saylor's zero-interest 2121/4242 leverage plans, and Coinbase joining the S&P 500. The bulk of the episode centers on Saudi Arabia's massive AI investment blitz (Amazon-Humain's $5B deal, Nvidia's 18,000-chip sale) during Trump's Riyadh trip, framed as a geopolitical realignment that locks the Gulf into the US AI camp against China, plus a debate over whether AI leadership will be winner-take-all or fast-democratizing. Other threads include the FDA's new AI-assisted drug review pilot, CoreWeave's IPO validating a compute-scarcity thesis, AI-driven coding disruption (Blitzy, Microsoft layoffs) and Nvidia's cheap domain-adaptive pretraining trick, DeepMind's self-improving AlphaEvolve, the $101M XPRIZE Healthspan milestone awards, and a closing segment on Elon Musk's simultaneous Saudi robotaxi/Starlink deals, Tesla's undervaluation, and the Boring Company's fully automated tunneling. A recurring macro theme ties collapsing global birth rates to the urgency behind humanoid robotics investment.

▶ Watch on YouTube

Topics

Bitcoin's bull run and MicroStrategy's debt-free leverage engine Crypto/Web3 ▶ 4:21
Bitcoin reclaims the $100K+ zone as Michael Saylor doubles down with new 2121/4242 equity-and-debt raises; the panel explains how removing interest payments makes MicroStrategy's Bitcoin-buying 'infinitely scalable' and discusses the hedge-fund arbitrage behind the zero-interest debt.
Coinbase joins the S&P 500 and crypto's mainstream legitimacy Crypto/Web3 ▶ 9:24
Coinbase's S&P 500 inclusion days after Bitcoin passed $100K is framed as a legitimacy tipping point, though wealth managers still bar advisors from recommending Bitcoin over legal-liability fears.
Saudi Arabia's AI investment blitz with US tech giants AI ▶ 16:08
Amazon's $5B AI-zone deal with Saudi's Humain and Nvidia's $10B, 18,000-GB300-chip sale anchor a week where US AI/finance leadership descended on Riyadh, cementing Saudi Arabia's alignment with US infrastructure over China.
AI dominance race: winner-take-all vs. fast democratization Geopolitics ▶ 27:59
Peter and Dave argue AI national leadership will be locked in within two-to-four years and that concentration forces are winning; Salim counters that history (OpenAI to Gemini to DeepSeek) shows capability democratizes too fast for a permanent winner.
Google vs. OpenAI vs. Nvidia: valuations and the chip stack war Compute ▶ 32:36
Discussion of Google ($2T) vs. OpenAI ($300B) vs. Nvidia ($3.3T) valuations, Google's efficient new TPU7 chips threatening Nvidia's dominance, and Google's search-revenue cannibalization risk from AI.
Taiwan/TSMC chip supply risk Geopolitics ▶ 34:56
China's stated goal of controlling Taiwan by 2028 and a federal planning assumption of zero Taiwan chip access by December 2027 are discussed as an existential bottleneck for the US AI buildout.
FDA's AI-assisted drug review pilot and deregulation push Health ▶ 35:34
The FDA's new Gen AI review tool rolling out across all centers by June 30 is praised as fixing regulatory asymmetry (only penalizing bad approvals, never delay deaths), alongside Florida's new in-state stem-cell legislation.
CoreWeave's IPO and the 'compute-bound forever' thesis Compute ▶ 39:07
CoreWeave's 400%+ post-IPO revenue growth is read as a bellwether reopening the IPO/venture market, resting on the bet that AI compute demand ('Computronium') will stay scarce indefinitely.
AI coding disruption, Blitzy, and preparing the workforce AI ▶ 42:08
Microsoft layoffs hitting engineers hardest (40% of cuts) frame a discussion of Blitzy's overnight full-codebase rewrites, the coming end of most white-collar coding jobs by 2030, and Dave's push for CEOs to plan humanely for displaced workers.
Self-improving AI: domain-adaptive pretraining and AlphaEvolve AI ▶ 56:09
Dave describes Nvidia's cheap technique for retraining/distilling open models like Llama into narrow specialists (e.g., circuit design) for ~1% extra cost, then connects it to DeepMind's AlphaEvolve, which used Gemini to improve Gemini's own training speed and GPU performance.
Archer Aviation named official LA28 Olympics air taxi provider Transportation ▶ 1:12:40
Archer's eVTOL air-taxi deal for the 2028 LA Olympics is treated as a hard deadline forcing FAA regulation and manufacturing to catch up, with Peter hoping Santa Monica Airport becomes a vertiport.
XPRIZE Healthspan $101M milestone awards Longevity ▶ 1:16:20
Peter recaps the Saudi-backed $101M Healthspan XPRIZE (623 registered teams), the first $10M in milestone awards to 40 teams, and its ~30x leverage of R&D spend versus prize money.
Declining birth rates driving humanoid robotics investment, plus Elon's Saudi deals Robotics ▶ 1:20:12
Collapsing fertility rates (South Korea 0.72, China 1.16) are framed as the real driver of China's AI/robotics push, tying into projected $1/hour humanoid robot costs and Elon Musk's Saudi robotaxi/Starlink announcements, Tesla's undervaluation, and the Boring Company's autonomous tunneling.

Predictions made

open Dave Blundin: If 1% of the Fortune 1000 put 1% of their treasury into Bitcoin, the price would reach $1 million per Bitcoin.
EP #? · · due: unspecified · ▶ watch
“if 1% of the Fortune 1000 put 1% of their treasury in Bitcoin, that would drive it to a million dollars of Bitcoin, which seems more and more likely”
Your call:
open Salim Ismail: A flood of corporate treasuries will announce Bitcoin holdings in the next six months, following MicroStrategy's and Nakamoto/KindlyMD's lead.
EP #? · · due: late 2025 · ▶ watch
“I think we'll see a flood of this type of announcement in the next six months as people go, we got to get into this”
Your call:
open Peter Diamandis: AI dominance among nations will be effectively decided within the next two to four years, and whoever leads will make the rules going forward.
EP #? · · due: 2027-2029 · ▶ watch
“we're going to see the AI... dominance play out over the next two to three years, two to four years... who's on top and who's second and third is unlikely to change after that”
Your call:
open Dave Blundin: The window to set the course of one's career/life around AI closes in about four years (by 2030); after that the value distribution will already be locked in.
EP #? · · due: 2030 · ▶ watch
“the safest bet by far is to run like hell toward your targets... you'll be very glad you didn't waste a millisecond between now and 2030”
Your call:
open Dave Blundin: Almost any white-collar coding or other knowledge job will no longer exist.
EP #? · · due: 2030 · ▶ watch
“almost any white collar coding or other job, it's not going to exist in 2030”
Your call:
open Dave Blundin: China intends to take control of Taiwan.
EP #? · · due: 2028 · ▶ watch
“So what does China said? They want to own Taiwan by 2028”
Your call:
open Dave Blundin: The US will have no access to Taiwan-manufactured chips if China takes Taiwan, and US fabs won't be ready in time.
EP #? · · due: December 2027 · ▶ watch
“the federal estimate... is that you have to make the assumption that no access to chips from Taiwan by December 2027, which is nowhere near enough time to get the US fabs up and running”
Your call:
open Elon Musk (clip): Humanoid robots will scale to tens of billions of units, potentially creating an economy ten times the size of today's global economy with 'universal high income.'
EP #? · · due: unspecified · ▶ watch
“my prediction actually for humanoid robots is that ultimately there will be... tens of billions... potentially we could have an economy 10 times the size of the current global economy where... no one wants for anything”
Your call:
open Peter Diamandis: eVTOL air taxi manufacturing and FAA regulation will be operational in time for Archer to serve as official air taxi provider for the LA28 Olympics.
EP #? · · due: 2028 · ▶ watch
“the 28 Olympics is when... manufacturing operations, FAA regulations all need to be in place”
Your call:
open Peter Diamandis: Santa Monica Airport will close and be converted into a vertiport for eVTOL air taxis.
EP #? · · due: 2028 · ▶ watch
“it's supposed to be closed down in 2028... hopefully what will happen is it will become a vertaport... for these to take off and land”
Your call:

Numbers that matter

Worth digging into

🕳️ Nvidia's domain-adaptive pretraining cookbook
Dave describes Nvidia publishing an openly available recipe for retraining/distilling open models like Llama into narrow specialists (e.g., circuit design) for roughly 1% extra training cost, yet almost no startups exploit it — a large, underused edge for anyone building a vertical AI product.
🕳️ AlphaEvolve and self-improving AI
DeepMind's AlphaEvolve already improved Gemini's own training speed by 23% and GPU performance by 32.5% via AI-evolved code, and Sam Altman reportedly treats internal code self-improvement as uniquely sensitive at OpenAI — a live, running self-improvement loop worth tracking closely.
🕳️ MicroStrategy's zero-interest 2121/4242 debt engine
Saylor restructured financing into ~$84B of combined equity/debt raises with no interest payments, removing prior existential risk and creating what the panel calls an infinitely scalable Bitcoin-buying machine, propped up by hedge funds buying the debt while shorting the stock.
🕳️ Saudi Arabia's coordinated AI dash (Humain, Amazon, Nvidia, Cerebras, CoreWeave)
Humain plus a wave of announced deals is a coordinated bid by Saudi Arabia to become the '#2 AI power' behind the US, deliberately aligning with US infrastructure over China — a fast-moving geopolitical realignment.
🕳️ Taiwan chip cutoff scenario (December 2027)
Dave cites a federal planning assumption that Taiwan chip access could end by December 2027 if China moves on Taiwan, with US fabs nowhere near ready — a concrete near-term risk to the entire AI buildout.
🕳️ Global birth-rate collapse as the real driver of robotics investment
The panel frames China's AI/robotics push as demographic survival rather than competition with the US, citing rates like South Korea's 0.72 and China's 1.16 against a 2.1 replacement rate — a macro thesis worth checking against hard population data.